3D object recognition: Representation and matching

  • Authors:
  • Anil K. Jain;Chitra Dorai

  • Affiliations:
  • Department of Computer Science, Michigan State University, East Lansing, Michigan 48824. jain@cps.msu.edu;IBM T.J. Watson Research Center, P.O. Box 704, Yorktown Heights, New York 10598. dorai@waston.ibm.com

  • Venue:
  • Statistics and Computing
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

Three-dimensional object recognition entails anumber of fundamental problems in computer vision: representation ofa 3D object, identification of the object from its image, estimationof its position and orientation, and registration of multiple viewsof the object for automatic model construction. This paper surveysthree of those topics, namely representation, matching, and poseestimation. It also presents an overview of the free-form surfacematching problem, and describes COSMOS, our framework forrepresenting and recognizing free-form objects. The COSMOSsystem recognizes arbitrarily curved 3D rigid objects from a singleview using dense surface data. We present both the theoreticalaspects and the experimental results of a prototype recognitionsystem based on COSMOS.